Optimized personalized management approach for moderate/severe OHSS: development and prospective validation of an OHSS risk assessment index

Abstract

Abstract STUDY QUESTION Can a simplified ovarian hyperstimulation syndrome (OHSS) risk assessment index be developed and validated with sufficient discrimination of moderate/severe OHSS from those without OHSS? SUMMARY ANSWER This easy-to-use OHSS risk assessment index shows good discriminative power and high calibration accuracy in internal and external validation cohorts. WHAT IS KNOWN ALREADY An early alert and risk stratification is critical to prevent the occurrence of OHSS. We have previously developed a multi-stage smartphone app-based prediction model to evaluate the risk of OHSS, but app use might not be so convenient in many primary institutions. A simplified OHSS risk assessment index has been required. STUDY DESIGN, SIZE, DURATION This training and internal validation of an OHSS risk assessment index used retrospective cohort data from January 2016 to December 2020. External validation was performed with a prospective cohort database from January 2021 to May 2022. There were 15 066 cycles in the training cohort, 6502 cycles in the internal validation cohort, and 8097 cycles in the external validation cohort. PARTICIPANTS/MATERIALS, SETTING, METHODS This study was performed in the reproductive medicine center of a tertiary hospital. Infertile women who underwent ovarian stimulation were included. Data were extracted from the local database with detailed medical records. A multi-stage risk assessment index was constructed at multiple stages. The first stage was before the initiation of ovarian stimulation, the second was before the ovulation trigger, the third was after oocyte retrieval, and the last stage was on the embryo transfer day if fresh embryo transfer was scheduled. MAIN RESULTS AND THE ROLE OF CHANCE We established a simplified multi-stage risk assessment index for moderate/severe OHSS, the performance of which was further evaluated with discrimination and calibration abilities in training and internal and external validation cohorts. The discrimination abilities of the OHSS risk assessment index were determined with C-statistics. C-statistics in training (Stages 1–4: 0.631, 0.692, 0.751, 0.788, respectively) and internal (Stages 1–4: 0.626, 0.642, 0.755, 0.771, respectively) and external validation (Stages 1–4: 0.668, 0.670, 0.754, 0.773, respectively) cohorts were all increased from Stage 1 to 3 with similar trends, and were comparable between Stages 3 and 4. Calibration plots showed high agreement between observed and predicted cases in all three cohorts. Incidences of OHSS based on diverse risk stratification (negligible risk, low risk, medium risk, and high risk) were 0%, 0.6%, 2.7%, and 8.3% in the training cohort, 0%, 0.6%, 3.3%, and 8.5% in the internal validation cohort, and 0.1%, 1.1%, 4.1%, and 7.2% in the external validation cohort. LIMITATIONS, REASONS FOR CAUTION The influence from clinical interventions including cryopreservation of all embryos cannot be eliminated and thus certain risk factors like estrogen level on trigger day might be assigned with a lower risk score. Another weakness of the study is that several preventive treatments, for instance oral aspirin and letrozole, were not recorded and evaluated in the model. Despite the robust reliability of OHSS assessment index, this tool cannot be used directly for clinical decision-making or as a diagnostic tool. Its value lies in its capacity to evaluate the prognosis of various interventions and to facilitate clinician–patient communication. The combination of this tool and further symptoms and examinations should be all taken into consideration for accurate and personalized management of OHSS. WIDER IMPLICATIONS OF THE FINDINGS The OHSS risk assessment index can be implemented to facilitate personalized counseling and management of OHSS. STUDY FUNDING/COMPETING INTEREST(S) This study was supported by National Key R&D Program of China (2022YFC2702504), Medical Research Fund Guangdong Provincial (A2024003), and Xinjiang Support Rural Science and Technology (Special Correspondent) Program in Guangdong Province (KTPYJ 2023014). All authors had nothing to disclose. TRIAL REGISTRATION NUMBER N/A.